Use existing feed as cache when present

lib/comic-page.php

@@ -6,9 +6,10 @@ class ComicPage {
     public string $pageNo = '';
     public string $title = '';
     public string $imageUrl = '';
+    public int    $pubDate = 0;
 
     function __construct() {}
     function __toString() {
-        return "[{$this->href}] [{$this->imageUrl}] {$this->pageNo} // {$this->title}";
+        return "[{$this->href}] [{$this->imageUrl}] {$this->pubDate} {$this->pageNo} // {$this->title}";
     }
 };

lib/fetch-site-content.php

@@ -7,9 +7,14 @@ include_once('lib/log.php');
 include_once('lib/fetch-url.php');
 include_once('lib/comic-page.php');
 
-function fetchSiteContent() {
+function fetchSiteContent($knownContent = array()) {
     $comicPages = array();
 
+    $knownPageUrls = array();
+    foreach ($knownContent as $knownItem) {
+        $knownPageUrls[$knownItem->href] = $knownItem;
+    };
+
     $archivePage = fetchUrl(Config::archiveUrl);
 
     foreach ($archivePage->find('div.textcontent ul') as $arc) {
@@ -17,9 +22,17 @@ function fetchSiteContent() {
 
         $comicLinks = $arc->find('a');
         foreach ($comicLinks as $el) {
+            $pageUrl = Config::baseUrl . "/" . $el->href;
+
+            if (@$knownPageUrls[$pageUrl]) {
+                Log::debug("Using cached version of " . $pageUrl);
+                array_push($comicPages, $knownPageUrls[$pageUrl]);
+                continue;
+            };
+
             $link = new ComicPage;
 
-            $link->href = Config::baseUrl . "/" . $el->href;
+            $link->href = $pageUrl;
             $link->pageNo = $el->innerText;
 
             // we'll get the image URL from this later, but fetch it early

lib/generate-feed.php

@@ -3,36 +3,45 @@ declare(strict_types = 1);
 
 include_once('config/default.php');
 
-function generateFeed($content) {
+function generateFeed($unsortedContent) {
     $loader = new \Twig\Loader\FilesystemLoader('templates/');
     $twig = new \Twig\Environment($loader, [
         'autoescape' => false,
         'strict_variables' => true
     ]);
 
-    $ts = time();
+    $fakeNow = time();
+    $realNow = time();
 
     $items = [];
 
-    foreach ($content as $page) {
+    $content = $unsortedContent;
-        // artificially enforce ordering on undated historical items
+    // Sort by date descending (newest first)
-        $ts -= Config::feed['itemDelaySeconds'];
+    usort($content, fn($a, $b) => $a->pubDate < $b->pubDate ? 1 : -1);
-        $now = date('c', $ts);
 
+    foreach ($content as $page) {
         $content = $twig->render('item-content.html', [
             'url' => $page->href,
             'imageUrl' => $page->imageUrl
         ]);
 
-        $items[] = $twig->render('item.xml', [
+        $item = $twig->render('item.xml', [
             'title' => htmlspecialchars($page->title),
             'pageNo' => $page->pageNo,
             'fullTitle' => htmlspecialchars($page->pageNo . " // " . $page->title),
-            'date' => $now,
+            'date' => $page->pubDate ? date('c', $page->pubDate) : date('c', $fakeNow),
             'url' => $page->href,
             'imageUrl' => htmlspecialchars($page->imageUrl),
             'content' => htmlspecialchars($content)
         ]);
+
+        $items[] = $item;
+
+        // we use this to artificially enforce ordering on undated
+        // historical items (but note that it starts with the real
+        // time, so that when we see a newly published item it's
+        // accurately dated)
+        $fakeNow -= Config::feed['itemDelaySeconds'];
     };
 
     $feed = $twig->render('feed.xml', [
@@ -40,7 +49,7 @@ function generateFeed($content) {
         'description' => Config::feed['description'],
         'baseUrl' => Config::baseUrl,
         'feedUrl' => Config::feed['feedUrl'],
-        'date' => $now,
+        'date' => $realNow,
         'items' => implode("\n", $items)
     ]);
 

lib/read-feed.php

@@ -25,10 +25,11 @@ function readFeed($path) {
         $item->pageNo = (string) $itemXml['data-pageno'];
         $item->title = (string) $itemXml['data-title'];
         $item->imageUrl = (string) $itemXml['data-imageurl'];
+        $item->pubDate = strtotime((string) $itemXml->pubDate);
 
-        print $item;
+        array_push($items, $item);
-        exit();
-
-        array_push($item);
     };
+
+    Log::info("Read " . count($items) . " items from " . $path);
+    return $items;
 };

rssmonster.php

@@ -15,7 +15,7 @@ include_once('lib/fetch-site-content.php');
 include_once('lib/generate-feed.php');
 
 $existingFeed = readFeed(Config::feedPath);
-$content = fetchSiteContent();
+$content = fetchSiteContent($existingFeed);
 $feedXml = generateFeed(array_reverse($content));
 
 Log::info("writing " . strlen($feedXml) . " bytes to " . Config::feedPath);